Digitizer for Application of Computers to Automatic Amino Acid Analysis W. L. Porter and E. A. Talley, Eastern Regional Research Laboratory," U. S. Department of Agriculture, Philadelphia, Pa. 191 18 the Tquantities of individual amino acids HE
MANUAL
CALCULATJON
from the charts produced by an automatic amino acid analyzer, especially in complrx natural mixtures, can be extremely tedious and time-consuming, To a t least' partially eliminat,ethesf. factoi,e, a system has been designed \\-hich convert's the millivoltage output of t,he three photocells of an analyzer, built to the specificat'ions of Spackman, lloore, and Stein ( d ) , to IBMb coded tape. This tape is used as the input for a properly programmed electronic computer. By this means, a complete basic amino acid analysis or a neutral and acidic amino acid analysis can be calculated in from 8 to 16 minutes, depending upon the kind and amount of information sought, with an accuracy equal t,o or better than that obtained by the manual procedure. The interest in a system for computer calculation, such as this, is quite widespread as evidenced by the number of inquiries received from those who have heard of it by various means. hlthough other approaches, somewhat different. from t.he one being described, are known to be under study both in research laboratories (1) and probably in instrument companies, this brief description is presented because of it's demonstrated prac t,icability. In the application described, the output signals of the analyzer photometer are in the form of analog signals. However, this is complicated by the use of three photocells the outputs of which must be monitored simultaneously. One represent's the amount of 570-mp light transmitted by the solut,ion, another represents the amount of 440mp light transmitted, and the third represents the amount of 570-mp light transmitted by the same solution but through a shorter path distance than the first. The dat,a collection system must accept each of these analog signals, convert them to digit,al forms, and record them on punched paper t'ape in the order given. In addition, each individual digital data point must be recorded at the identical time its corresponding analog point is being recorded on the analyzer recording chart. This is necessary because two recording potent,iometers, when used in parallel, must balance a t the same time if accurate values are to be obtained. (Potentiometers tend to draw current until they balance and 1692
ANALYTICAL CHEMISTRY
assemble the components required for this applicat,ion. However, several accessory items are necessary to relat,e the digitizer to the automatic analyzer and to make the tape completely compatible with t,he paper tape reader on the computer.
Of
' .
Position
Programmer Control Ponel
Figure 1. Diagram digital converter
Punch
of
analog
to
thus reduce the voltage measured.) -Also this allows t,he operator easily to det'ermine the peak volumes for each amino acid to be used by the computer for identification. Use of the analog chart also allows the operator to determine the success or failure of a run. Simple use requirement specifications were drawn up and submitted for bids by electronic equipment manufacturers. The successful bidder then produced a combination of modules required to fulfill the specifications. System Description. A block diagram of the analog to digital converter and punch is given in Figure 1. The voltage indicating potentiometer receives the 0- to 10-mv. analog signal from the amino acid analyzer and rot,ates the shaft position encoder to a corresponding position. This encoder provides a direct digital output corresponding to the angular position of the indicat,or shaft. The control chassis stores and translates the digitally coded output of the encoder t,odecimally coded contact closures. The programmer provides controls to perform system functions according to a prearranged format. The control panel provides circuit breakers for system power, selects operating mode, and cont,rols record and reset functions. The power supply provides the d.c. power and amperage required for operation of the' other components. The tape punch punches the paper tape according to the commands received from the programmer.
The above description, although general in nature, provides sufficient information for a manufacturer to build or
Accessory Equipment. .A threeplace st'epping switch was mounted on the same shaft as the stepping switch in the analyzer recorder. Only the 570nip full light path photocell terminal is connected to the digitizer to give switch closure just prior to the 570-mp full light path photocell daba point.. This signal is fed to a counter which records seven data points from this photocell whkh correspond to 21 data points (63 digits) for the three photocells, and then command the digitizer to punch t'he end of line code. This is required to orient the computer with respect to the sequence of points if some mechanical or electrical trouble is encountered. This number of data points was chosen because these points can be conveniently read into the computer memory as a unit'. Therefore, both tape quantity and computer time are saved. h tandem microswitch was inst,alled on the analyzer recorder to produce a contact closure when a dot on the chart is being printed. This signals t,he converter to read the data and punch them into the tape for the reasons discussed above. The input leads to the converter were connected to the input, terminals of t'he analog recorder amplifier between the amplifier and the recorder stepping switch. In this way the photocells were connected, in turn, to both potentiometers in the same n a y a t the same time. In this case the converter indicating potentiometer has a onesecond balance time compared to five seconds for the analyzer multipoint recorder. The output of the converter is an eight-channel tape which, in this case, is punched in code capable of being read by the tape reader for an IBRl 1620 computer. In addition to the digit values (three digits per analog record) in millivolts, a series of tape feed punches (channels 1 t'hroiigh 7 ) is placed a t each end of the digital record to facilitate use with the computer. A switch on the converter is deprcssed a Eastern Utilization Research and Development Division, Agricult~uralResearch Service, U. S. Department of Agrirulture. Mention of commercial items does not carry its recommendation over any other not mentioned.
for a sufficient lengt'h of time to give the required length of tape feed. In the I B I I coding system all digits must have an odd number of holes in the talle. An even number of holes cannot be handled by the computer. parity checking auxiliary attachment was added to the punrh to indicate when an even number of holes was punched due to some difficulty in the converter rnechankm so that repair could be accwmplished in a minimum of time. This is especially important when considerable time elapses bet'ween the t'ape punching and the actual computer comput at ion. Sample information, such as column niimher and sample number, is written on each end of the tape for identification purposes. The computer program requires the effluent volume of each peak and the identification of the amino acid responsible. This information is obtained from the analog record (chart) and is ent,ered in the appropriat'e place on a line of an 80-column work sheet along with such items as column num-
ber, run number, tape start time, tot'al solids, and size of sample. This information is punched from the work sheet into a punch card which is fed into the computer memory to be used when called upon by the comput'ation program. The elect,roniccomput,ations are made a t the U.S. Department of Agriculture Research Service, Biometrics Division, Beltsville, LId. Information about the machine language program can be obtained from that organization if required. Performance. T h e instrumentation described has been in use for about 2 years, operating from 5 to 7 days per week. T h e percentage of lost time due to breakdown of components has been less t h a n 1%. T h e calculated d a t a have been demonstrated to have a n accuracy equal to, or better t h a n , t h a t obtained by the manual calculation from t h e corresponding analog chart. The higher accuracy results can probably he accounted for through the fact that all data points are employed in the computer calculation of peak areas while only selected points are used in the manual system. Use of this procedure
has eliminated the need for one professional chemist per year on the project. ACKNOWLEDGMENT
The authors gratefully acknowledge the help of the Datex Corp., Ilonrovia, Calif., who manufactured the equipment used. I n addition, they acknowledge the help of several engineers from the Leeds and Korthrup Co., Philadelphia, Pa., and of engineers from the International Business Machine Co., Philadelphia, Pa., in instruction in the basic electronics necessary for the final dwigning of the equipment. In addition, it is known that several other electronic manufacturers can assemble suitable equipment for use in t'his application; the United States Government cannot recommend any one of these instruments over another. LITERATURE CITED
(1) Moore,
S., private communication (1962). ( 2 ) Spackman, D. H., Stein, R. H., hloore, S., ANAL.CHEM.30, 1190-1206
(1958).
Apparatus for Quantitative Determination of Fluorine in a Mixture of Oxygen and Fluorine Stanley W. Comer, Air Force Rocket Propulsion laborafory, Research & Technology Division, Air Force Systems Command, Edwards, Calif. ~ T N T H E S I Sof fluoroxi compounds involves the analysis of mixtures of oxygen and fluorine. The apparatus described in this article was designed and built to determine the amount of fluorine in a mixture of oxygen and fluorine. THE
R B
A
B
STIRRINC 0 & R
EXPERIMENTAL
The apparatus is shown in Figure 1. Pressure d h i n the apparatus does not differ greatly from at'mospheric pressure; therefore, vacuum stopcocks are not, used. The stopcocks are lubricated with Kel-F No. 90 grease (Minnesota Mining & Manufacturing Co., St. Paul 6, Ninn.). To perform an analysis. the apparatus is filled with gascous oxygen and fluorine. Mercury is added to react with the fluorine. As a result of the reaction, pressure in the apparatus is reduced. After completion of the reaction, Kel-F polymer oil is added to restore atmospheric pressure in t'he api)aratus. The volume of oil plus t'he voliime of mercury equals the volume of fluorine in the sample. Opcration of the apparatus is as follows. Part .4 is connected to part B , and the apparatus is put into a constant temllerature water bath which isq)laced over a magnetic stirring motor. Kel-F oil (or concentrated sulfuric acid) is carefully poured into the manometer. Tube B is connected to a vacuum manifold which has a pressure gauge
C
PART A
Figure 1 .
PART B
MANOMETER
Fluorine analyzer
calibrated in millimeters of Hg. The manometer is isolated from the manifold by t i m i n g the three-way stopcock S I to the position which connects the manifold to the reaction flask. K h e n a pressure of 1 mm. or less is reached, stopcock S3 is closed. .A mixture of gaseous O2 and F2 is allowed to enter through tube .-I until the pressure in the apparatus is approximately 20 mm. above atmospheric pressure. Stopcock SI is closed and the temperature is allowed to stabilize for ten minutes. After the temperature has stabilized, excess pressure is relieved through stopcock S2. An excess amount of reagent grade mercury is slowly added through tube R. If mercury is added t'oo rapidly, gas may escape as stopcock S2is opened. The reaction between fluorine and mercury is complete when fresh mercury exposed by the stirring bar does not become discolored. Atmospheric pressure is restored in
the alqxiratus by adding Kel-F oi through tube R. Tube R i h initially filled iyith oil to a mark near the t o p of the reservoir. The amount of oil delivered is obtained by refilling the reservoir to the mark with oil from a buret. I t may be necessary to refill the reservoir several times. After adding a volume of oil equal to the estimated volume of fluorine in the sample minus the volume of mercury, stopcock SBis slon-ly turned to connect the reaction flask to the manometer. The volume of oil to be added or subtracted (arithmetically) from the amount, already in the flask is determined by the relative position of the t,wo nienihci in thc manonwtel~.which is calibrated in millilitcrs of oil. If the initial and final trniperatures and the barometric pressiire have not changed, the vohime of Kel-F oil plus the volume of mercury is equal to the volume of fluorine in the sample. RESULTS A N D DISCUSSION
A mixture of fluorine and oxygen was made by adding. fluorine to a large volume of oxygen until the mixture contained approximately 30y0 (by volume) fluorine as determined by pressure gauge readings. Results of analysis of the mixture by the procedure described in this article are given in Table I , Tests 1, 2 , and 3. Temperature of the water bath is givrn in VOL. 3 6 , NO. 8, JULY 1964
1693
